Randomized trial evaluates ARCHER's effectiveness in microarchitectural side-channel defense, highlighting its efficiency and configurations.
Microarchitectural side-channels leak data through caches, branch predictors, store buffers, and speculative execution. Evaluating defenses at cycle-model fidelity has forced a choice between functional tools (Spike, QEMU) and gem5’s hours-long Linux boots. ARCHER is a cycle-model-relative RV64IMAFDC RISC-V emulator with pluggable superscalar out-of-order execution, coherent caches, simultaneous multithreading, 1 to 16 harts, and thirteen speculation policies: an unrestricted baseline plus five classical defenses and seven new defenses spanning issue-gate, predictive-redirect, and selective-cleanup mechanism families. It boots Linux 6.6 in under three minutes, runs 3.7× faster (geometric mean) than gem5’s DerivO3CPU on head-to-head bare-metal microbench, matches gem5 within ±20% on 4 of 10 workloads, matches Chipyard’s fab-ready RTL to a 9.4% mean cycle-count deviation under two fitted match-configurations, and produces bit-identical architectural output to QEMU at a median 88× host-throughput advantage over Chipyard’s Verilator flow. Every published Spectre-family defense drives leakage to zero at sub-0.3% instructions-per-cycle (IPC) loss on a full Linux boot, and each new policy is validated on the attack surface its mechanism defends. The Adaptive Memory Encryption Scheme (AMES) adds a bus-layer engine with four authenticated ciphers (each validated bit-exact against its published specification) and per-leaf mask-XOR for Differential Power Analysis (DPA) hardening; a first-order Correlation Power Analysis consistency check confirms the direction of the theoretical bound under the shipped leakage model. Together, ARCHER and AMES provide a single INI-configurable environment for cycle-model side-channel evaluation in minutes rather than hours.
No takes yet. Share an insight, caveat, or question.
Mishra et al. (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: